Bertrand Lallau [Sat, 1 Feb 2014 13:33:33 +0000 (14:33 +0100)]
Remove a catching exception during delete_volume
GlanceMetadataNotFound Exception was catching during
a call to volume_glance_metadata_delete_by_volume().
This exception is never thrown by the implementation.
This catching has been removed.
Ryan McNair [Mon, 27 Jan 2014 21:48:51 +0000 (21:48 +0000)]
Add support for special char in volume metadata
Using special characters such as "&" in volume metadata caused SAX
parser to split text into multiple child nodes, as stated in the
specs for SAX parsers. In this case extract_text would return
'None' because text was return if there was exactly
one child node. Concatenating the values of child nodes accounts
for SAX parsers splitting contigous characters into
multiple chunks.
Brianna Poulos [Wed, 29 Jan 2014 20:49:20 +0000 (15:49 -0500)]
Stop volume_type_encryption creation when in use
This bug fix addresses bug #1274252. It requires a volume
type to have no volumes with the type before allowing it
to be made encrypted. This prevents a situation where
non-encrypted volumes could have an encrypted volume type.
This fix also adds a unit test to confirm functionality.
john-griffith [Wed, 29 Jan 2014 04:30:59 +0000 (21:30 -0700)]
Revert initialize_connection changes
A change was merged that reusulted in interittent
gate failures. It seems that in certain cases the
intitialize is doing an update targets and is
deleting the existing targets and not recreating them.
The idea of this patch was to fix up the iscis targets
for cases where volumes are extended.
A better solution is in progress that will remove the
target creation from the create_volume process altogether.
Currently we seem to have create_export, ensure_export and
now initialize that all are designed to do very similar
things.
A bug has been filed to address this and will attempt
to collapse these functions to be done at attach time.
Florian Haas [Mon, 20 Jan 2014 21:42:11 +0000 (22:42 +0100)]
VolumeManager: initialize even if a volume can't be found
If a previously volume cannot be exported (for example, because
an iSCSI target's backing LV is no longer present), VolumeManager
would previously bail out, leaving the volume service uninitialized.
This left volumes dangling in limbo: their state would be reported
as available (clearly not a reflection of reality), and they could
not be deleted (because the volume service that would be responsible
for deletion was unavailable).
Instead, catch an exception raised by ensure_export() separately,
log it, and set the volume to the error state.
For the tgt backend, this will also remove the volume_path file.
Previously this would cause an ISCSITargetRemoveFailed exception
when the volume would be removed. Instead, simply log a warning
and return, the way other backends (example: RBD) already do.
Also, add an info message that reflects the actual path and
contents of the volume_path file.
Finally, fix up the tgtadm unit test so that it tests for
identical IQNs,
Eric Harney [Mon, 27 Jan 2014 23:46:26 +0000 (18:46 -0500)]
Add create_iscsi_target stub to TargetAdmin
This should be defined in TargetAdmin, as it is assumed to be
present in all subclasses by callers.
IetAdm and LioAdm may need additional work to behave as desired
in all circumstances (such as volume extend), but this should
at least stabilize things in the meantime.
This patch validates the keys of the extra specs
before setting them.
This will make possible to remove those keys.
It allows alphanumeric characters, underscores,
periods, colons and hyphens.
Zhiteng Huang [Thu, 9 Jan 2014 06:54:22 +0000 (14:54 +0800)]
Make sure report_interval is less than service_down_time
Services that inherit service.py/Service class would register
themselves to DB and then update stats periodically (every
report_interval second). The consumer of this kind of information,
like scheduler or 'os-service' API extension, will consider a service
is 'up' (active) if last update from that service is not longer than
'service_down_time' ago.
The problem is if 'report_interval' was configured/provided greater
than 'service_down_time' by mistake, services would then be always
considered in 'down' state, which can result in unsuccesful placement
of volume create request for example. This is what Bug #1255685 is
about.
In previous fix: https://review.openstack.org/#/c/60760/, a
configuration check helper function basic_config_check() was added
*wrongly* to WSGIService class instead of Service class. This patch
moves the configuration check helper function and the check to the
right place to make sure 'report_interval' is less then
'service_down_time'.
john-griffith [Sat, 25 Jan 2014 02:12:01 +0000 (19:12 -0700)]
Fix up the way we do iqn variable in unit test
A recent bug was reported when a developer was trying to work
with the iscsi target unit tests. There were two problems discovered,
one of which has been addressed by the dependent patch of this commit
(https://review.openstack.org/#/c/68223/) which fixed the inconsistencies
in the string we used for iqn's in our test.
This patch addresses another issue that's somewhat related and was
exposed when the developer tried to use the "self.target_name" variable
instead of hard-coding the iqn's in the tests.
The problem is that the target_name variable uses an invalid format
for the iqn based on cinder-defaults. It neglected to use the
volume_name prefix (in this case: "volume-").
This patch modifies the self.target_name variable to be more accurate
with what we would see in real usage, it also modifies the test to just
use this variable rather than hard coding it and running the risk of
having mismatches again in the future.
John Griffith [Fri, 24 Jan 2014 21:51:21 +0000 (14:51 -0700)]
Catch new iscsi exception
A while back we fixed up a target create/update error
that was identified in the gating tests:
https://review.openstack.org/#/c/47513/
Since then a new patch was introduced to to the update and separate
it so that it could easily be used in other places.
https://review.openstack.org/#/c/58599/
The problem is this added a new exception "TargetUpdate" and
didn't add this to the except block that we have.
This patch changes the catch block to handle the new exception and
do the retry, thereby putting the original fix back in place.
ling-yun [Mon, 20 Jan 2014 10:48:43 +0000 (18:48 +0800)]
Delete volume transfer in volume_destroy function
Doing volume force delete operation on volume-A will change volume-A's
status to 'deleted', so all related data about volume-A should be deleted.
Now current version's volume_destroy function doesn't delete related data
of volume transfer. If there is volume transfer about volume-A, first do
force delete operation on volume A, and thus the volume-A's transfer data
would be residual. So remove the related data of volume transfer in
volume_destroy function.
Zhiteng Huang [Wed, 8 Jan 2014 06:50:29 +0000 (14:50 +0800)]
Enable multi-process for API service
Due to the limit of Python interpreter, API service of Cinder can't
really utilize underlying multi-core architecture even libraries
like eventlet has been used. To make API service much more scalable,
we'd adopt multi-process (worker) mode that has been used for long
in Glance/Swift/Nova.
The default behavior isn't changed with this patch, Cinder API
service will still run in one process (default value of
osapi_volume_workers is None).
Implementation wise, a good portion of cinder/service.py has been
removed because those content has been merged in Oslo version of
service module. cinder/wsgi.py is also updated to adopt the change
for multiple WSGI servers running in separate processes.
Implement bp: multi-process-api-service
DocImpact: 'New config option osapi_volume_workers is used to specify
number of API service workers (OS processes) to launch for Cinder
API service. Setting this config option to a proper value (e.g.
osapi_volume_workers = # of CPU cores/threads of the machine) can
greatly improve the total throughput of API service [# of API
requests can be handled per second].'
Also removed out-dated comments in bin/cinder-api due to the fact
that this bug [1] has been fixed in eventlet 0.9.13
Eric Harney [Thu, 16 Jan 2014 19:08:53 +0000 (14:08 -0500)]
Sync common modules from Oslo
Identified as dependencies for RPC sync by update.py
Contains code from: 12bcdb7 Remove vim header e6494c2 Use six.iteritems to make dict work on Python2/3 6d0a6c3 Correct invalid docstrings fe18bfa Remove keystone from default_log_levels default 0228b81 Logging excepthook: print exception info if debug=True 8b2b0b7 Use hacking import_exceptions for gettextutils._ 6d0a6c3 Correct invalid docstrings fe18bfa Remove keystone from default_log_levels default 13eb01c Adding domain to context and log 27b21bc Unify different names between Python2/3 with six.moves a6f40ac Don't log to stdout when log_dir is set 4c22556 Use py3kcompat urlutils functions instead of urlparse 0bf03b7 Add network_utils.urlsplit e456727 Remove useless logging in networks_utils 8b2b0b7 Use hacking import_exceptions for gettextutils._ 37e46bb disable SIGHUP restart behavior in foreground f08f6ba service: replace eventlet event by threading 32cd4c0 Allow configurable ProcessLauncher liveness check 25ff65e Make wait & stop methods work on all threads 7013471 Fix typo in parameter documentation (timeutils) ae931a9 Avoid TypeError in is_older_than, is_newer_than ad17a69 Fix filter() usage due to python 3 compability
Oslo version: 7a51572 Merge "Implement cache abstraction layer"
Date: Wed Jan 15 19:31:16 2014 +0000
Eric Harney [Thu, 16 Jan 2014 18:25:33 +0000 (13:25 -0500)]
Sync py3kcompat, sslutils, versionutils from Oslo
Dependencies identified by update.py for RPC update.
py3kcompat: 12d3bbc Add method quote_plus in module py3kcompat.urlutils 8575d87 Removed copyright from empty files 0d8f18b Use urlutils functions instead of urllib/urllib2 12bcdb7 Remove vim header 4c22556 Use py3kcompat urlutils functions instead of urlparse 71743d9 Add urlopen/URLError/pathname2url in urlutils 0f2906f py3kcompat: Add unquote eaec23b py3kcompat: Add urljoin compatibility 35c317b Fix capitalization, it's OpenStack 4d12ad1 python3: Import translation layer
sslutils: 8b2b0b7 Use hacking import_exceptions for gettextutils._ 12bcdb7 Remove vim header 1a2df89 Enable H302 hacking check 99b7c35 Convert kombu SSL version string into integer b0b37ce Fix IBM copyright strings f63ea05 Use oslo-config-2013.1b3 e50b68c Support for SSL in wsgi.Service
versionutils: 8b2b0b7 Use hacking import_exceptions for gettextutils._ a5ae087 fixed typos 45658e2 Fix violations of H302:import only modules 37ea814 Adds decorator to deprecate functions and methods 12bcdb7 Remove vim header d7d74a7 Add `versionutils` for version compatibility checks
Oslo version: 7a51572 Merge "Implement cache abstraction layer"
Date: Wed Jan 15 19:31:16 2014 +0000
Eric Harney [Thu, 16 Jan 2014 18:09:05 +0000 (13:09 -0500)]
Sync gettextutils from Oslo
Update gettextutils as a dependency for RPC updates
Requires changes within Cinder as well for
"get_localized_message" -> "translate" method rename.
221c37d Allow the Message class to have non-English default locales 0e1af5b Implementation of translation log handler f5686b1 Merge "Translation Message improvements" 8b2b0b7 Use hacking import_exceptions for gettextutils._ 2cfc1a7 Translation Message improvements 45658e2 Fix violations of H302:import only modules 12bcdb7 Remove vim header
Oslo version: 7a51572 Merge "Implement cache abstraction layer"
Date: Wed Jan 15 19:31:16 2014 +0000
Avishay Traeger [Wed, 22 Jan 2014 12:27:39 +0000 (14:27 +0200)]
Storwize driver cleanup
1. Moved Storwize driver into 'ibm' directory and refactored into
separate files to improve readability and debug-ability. Other IBM
drivers will be moved into this directory in a separate patch.
2. Errors coming from storage now raise VolumeBackendAPIException,
and those coming from the driver raise VolumeDriverException.
3. Fixed many pylint errors.
4. Removed any usage of mox from unit tests in favor of mock.
Li Min Liu [Thu, 9 Jan 2014 06:54:09 +0000 (14:54 +0800)]
Add support for retype in Storwize/SVC driver
In Storwize/SVC, retype can be handled by a few cases:
1. The change may just require setting an option on the volume
2. The change may require moving the data in the same pool (e.g.,
compression)
3. The scheduler may have chosen a different pool virtualized by the
same Storwize or SVC, in which case we move the data as in #2
Brick's TargetAdmin classes have a
default value set in __init__ for
iscsi_target_prefix. The unit test was changing
the value returned for the internal _get_target,
but wasn't changing the CONF.iscsi_target_prefix to
match. So the default value in the __init__ was
being used and hence the differences.
This patch sets the configuration's iscsi_target_prefix
which alters the values passed into the driver. This
also makes the tests consistent between the iqn of the volume
at create time and delete time.
KIYOHIRO ADACHI [Fri, 10 Jan 2014 09:22:36 +0000 (18:22 +0900)]
Remove unused exceptions
The following exceptions are not used.
class MigrationNotFound(NotFound):
class BadDriverResponseStatus(VolumeDriverException):
class InstanceNotFound(NotFound):
class VolumeRetypeFailed(CinderException):
class DuplicateSolidFireVolumeNames(SolidFireDriverException):
Fix up calculating space info for mirrored volumes
The status reporting in the lvm driver for cinder volume incorrectly
reports free capacity. This is due to a couple of issues:
a) the number of mirrors isn't taken into account
b) there are some pathological cases where a
simple division by total mirrors isn't sufficient
because all mirrors must be on separate physical volumes.
Code Review / openstack-build / cinder-build.git / log